E3S Web of Conferences 107, 02001 (2019) https://doi.org/10.

1051/e3sconf/201910702001
ICSREE 2019

Fault detection and diagnosis of photovoltaic system using fuzzy

logic control
Sayed A. Zaki1,3,*, Honglu Zhu1,2 , Jianxi Yao1,2
1
School of Renewable and Clean Energy, North China Electric Power University, 102206, Beijing, China
2
The State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources, North China Electric Power
University, 102206, Beijing, China.
3
Faculty of Engineering, Cairo University, Giza, Egypt

Abstract. Among several renewable energy resources, Solar has great potential to solve the world’s energy
problems. With the rapid expansion and installation of PV system worldwide, fault detection and diagnosis
has become the most significant issue in order to raise the system efficiency and reduce the maintenance
cost as well as repair time. This paper presented a method for monitoring, identifying, and detecting
different faults in PV array. This method is built based on comparing the measured electrical parameters
with its theoretical parameters in case of normal and faulty conditions of PV array. For this purpose, three
ratios of open circuit voltage, current, and voltage are obtained with their associated limits in order to detect
eight different faults. Moreover, the fuzzy logic control FLC method is performed for studying the failure
configuration and categorizing correctly the different faults occurred. The outcomes obtained by performing
the different faults representing permanent and temporary faults demonstrated that the FLC was equipped to
precisely identify the faults upon their occurring. Different simulated and experimental tests are conducted
to demonstrate the performance of the proposed method.

1 Introduction technology-based monitoring by calculating Voltage and

Current Ratios VR, and IR is used to detect the different
Over the last few decades, the use of PV systems has faults. Some other techniques such as in ref. [7-8] are
spread rapidly all over the world in different applications, implemented based on the real-time environmental
from space to the residential, commercial, and industrial conditions and some PV parameters, however, the
applications [1]. High use of PV panels has brought high drawback of this methods is the high cost of equipment.
PV component failure rate as well. This high rate is Moreover, Partial Shading PS in the GCPV plant using
because of many reasons such as; short circuit between statistical methods is discussed and validated using
modules; an open circuit in different strings, etc. These Ratios VR& PR [9-10].
failures greatly influence on the operating efficiency of Recently, a number of fault diagnosis methods based
the PV generation systems and the performance of the PV on artificial intelligence techniques are adopted to
system work. Therefore, many types of research and increase the detection rate and right classification such as
developments in ref. [2] made good progress using on- Fuzzy Logic Control FLC [7, 10], Neural Network [2,11],
line fault diagnosis to increase system reliability and both Neural and FLC [10,12], Genetic Algorithm [13], or
performance by early fault detection. MATLAB [4, 6]. Currently, the FLC is widely used with
Different faults can be categorized into temporary and GCPV systems, Such as [10, 12] which presented FLC to
permanent faults in PV arrays, and both are responsible detect the faulty modules, but couldn't discriminate
for reducing the output power and solar energy generated between open circuit and short circuit faults.
compared with healthy operating conditions [3]. The work presented in this paper is to introduce a new
Although the temporary faults occurred for a short period technique to detect and diagnose the different types of
in reality, the protection devices need to discriminate this faults occurring in the PV power plants using Sugeno
condition from permanent fault conditions for preventing FLC method. This method is used as an artificial method
wrong shutdowns [4]. Therefore, to reduce the to increase the accuracy of fault identification and fast
unavailability time, increase the stability and efficiency diagnosis. The proposed detection method depends on the
of PV systems, different fault diagnosis and detection analysis set of three input ratios which are Open Circuit
techniques are needed to ensure the continuity of solar Ratio OCR, Current Ratio IR, and Voltage Ratio VR.
generation [5]. Calculating these ratios is performed using the measured
There are many fault detection methods used in the and simulated under normal and faulty conditions. The
GCPV systems. In [6], automatic supervision by OPC fault detection method is implemented using FLC-based

* Corresponding author: eng_sayed_002010@yahoo.com

© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution
License 4.0 (http://creativecommons.org/licenses/by/4.0/).
E3S Web of Conferences 107, 02001 (2019) https://doi.org/10.1051/e3sconf/201910702001
ICSREE 2019

method and MATLAB Simulink tool to identify the type

of the fault. The selected faults in this work are PS, PS
with bypass diode failure, open circuit, short circuit,
snow falling, and bird or tree leaves dropping on PV
plant. The algorithm is validated using 3.34 kWp solar
PV system installed at the roof of North China Electric
Power University (NCEPU). Fig. 1. The single diode model of the PV module.
The paper is organized as follows: the next section
presents the different faults types in a PV plant including
the solar PV modeling. Then, the methodology of the 2.2 Faults of PV array and the output features
work is discussed in Section 3. Section 4 discusses
Faults in the PV system not only decrease output power
Sugeno FLC implementation. In section 5, the
generation, but it also reduces the availability and
experimental test results, discussion, and comparison
reliability of the whole system. Therefore, in order to
with other methods are provided. Section 6 provides
maximize the output power, identifying the faults in the
conclusion remarks.
PV system is extremely important. There are many faults
affecting on the DC side the PV system temporarily or
2 Different faults in PV array may be permanent.
This work aims to identify, detect, and localize the
In this section, the modeling of PV array is presented by faults occurring in the PV array. The selected faults in
defining the mathematical equations, then the different this work are discussed in Table 1. In addition, the
faults may occurring in PV system and their MATLAB Simulink model is built to extract the different
characteristics are investigated in detail to extract the features of these faults from I-V, and P-V curves as
different features of every fault case. shown in the following section.
Table 1. The different selected faults occurred in the PV system
2.1 Solar PV model
Fault type Symbol
The most commonly utilized models of PV module Normal operation F0
essentially incorporate the one diode (or called the five- PS effect without BP failure F1
parameters model), the two-diode, the four-parameter PS effect with BP failure F2
model, etc. They can be utilized generally for various Two modules short circuit in PV array F3
application events or distinctive purposes. In this paper, Four modules short circuit in PV array F4
the one-diode model is utilized as a mathematical model Two modules open circuit without BP
F5
for obtaining the relation between the module current and failure in modules
voltage. Then, the DC side of the PV system is modeled Four modules open circuit without BP
F6
using Matlab Simulink. The voltage and current Values failure in modules
Snow falling F7
of the PV panel can be calculated as shown in Fig. 1,
Bird or tree leaves dropping F8
which can be expressed as [11]:
(1)
2.2.1 Partial Shading PS fault without bypass diode
( ) (2) failure (F1)

Where is the light-generated current at Standard The shading profiles can be extremely unexpected due to
Test Condition STC, is the diode current, is the non-uniform of radiation during occurring that condition.
current in shunt resistance, is the dark saturation The all PV array is assumed shaded with solar radiation
current at STC, is the panel series resistance, is G= 500 W/m2 at 25 oC of temperature T at Standard Test
the panel shunt resistance, is the number of series cells Condition STC ( G=1000 W/m2, and T= 25 oC ).The -
in the PV panel and is the thermal voltage and it can and 𝑃𝑃- characteristics are illustrated in Fig. 2, for
be defined based on: normal and fault cases analysis. In case of PS situation,
the short circuit current remains constant, but the
(3) open-circuit voltage significantly reduces with the
Where the ideal diode factor, is Boltzmann's increase of the shaded modules.
constant and is the charge of the electron.
Although, the environmental conditions of solar 2.2.2 Partial shading fault with bypass diode failure
radiation and temperature have a direct influence on the (F2)
PV modules output characteristics. Therefore, their effect
should be taken into consideration during studying the The assumption of this case is all bypass diodes that
different features in case of normal and fault cases. connected parallel with the PV modules are open
circuited during the case of PS occurring. The received
solar radiation density is assumed to be G= 500 W/m2 at
25 oC. - and 𝑃𝑃- characteristics of this case cases are
shown in Fig. 2. In this case, the maximum power 𝑃𝑃 ,

2
E3S Web of Conferences 107, 02001 (2019) https://doi.org/10.1051/e3sconf/201910702001
ICSREE 2019

and of the PV system significantly decrease due to the in Fig. 2, and 𝑃𝑃 dramatically decrease due to
voltage drop lost on the shaded modules while bypass reduction in the solar radiation, while reduces
diode not working properly, while rests the same of significantly with the rise in the number of covered
previous PS. modules by snow.

2.2.3 The short circuit fault at STC (F3, and F4) 2.2.6 Bird or tree leaves dropping at STC (F8)
The short-circuit fault is an accidental interconnection in This kind of temporary fault occurs due to presence of
the current carrying conductor, PV cells/modules bird or leaves prevent totally the solar radiation to be
cracking, or connections between modules with received by the PV module. It's assumed that one full
considerable effect on the series and parallel resistance of module is fully covered, i.e. the solar radiation equals 0
the PV array. In this part, it's assumed in F3 and F4 that W/m2 at a normal temperature, and the remaining PV
PV modules have a short circuit problem in two modules, modules receives the normal solar radiation. As shown in
and four modules respectively. From shown in Fig. 2, Fig. 2, is the same, while and 𝑃𝑃 significantly
nearly the same, while and 𝑃𝑃 of these cases reduce with the rise in the numbers of covered modules.
significantly reduce with the rise in the numbers of For detecting these kinds of faults, a method of
shorted modules. comparing the ratios with their threshold high and low
limits is discussed and implemented using FLC algorithm.
2.2.4 Open circuit fault at STC (F5, and F6)
3 Methodology of PV faults diagnosis
The open-circuit fault is an unexpected disconnection
inside the PV module conductor which influence on the and detection algorithm
array total resistance during the fault. In this part, we The fault diagnosis technique used in this paper is based
assume in F5 and F6 that PV modules have a on the threshold limits, furthermore, one normal and eight
disconnection problem in two modules and four modules, faulty categories are discussed and detected, then, the
respectively while the bypass diode is connected in ratios between the measured and the simulated
parallel with the modules. - and 𝑃𝑃- curves have been parameters of PV string are compared with threshold
compared with the PV array in case of short circuit case, limits in order to detect the exact type of fault. Moreover,
as shown in Fig. 2. almost remains constant, while the analysis of the main attributes in the I-V
and 𝑃𝑃 of these cases slightly drop with the increase of characteristics of each string in the PV array plays the
the disconnected module numbers due to voltage drop on main role to localize faults accurately.
the bypass diode equals 0.8 v across each faulty module, Three ratios have been identified to detect different
which directly has a significant influence on the total types of faults in DC side; , , and have been
output power, and voltage. This means power gets lost in used to categorize the region where the fault has occurred.
bypass diode which is given by [13]: All ratios can be calculated using the following
𝑃𝑃 (4) expressions:
Where, is the module output current and is
(5)
the voltage drop across the bypass diode.
(6)

2.2.5 Snow cover fault (F7) (7)

where , , are the theoretical
This fault occurs due to snow falling only in winter as a parameters of , , and generated by the PV system
temporary fault causes large energy losses for system MATLAB model respectively, while , ,
operation. Assume in this fault, the all PV array only and represent the measured data of same
receives 10% solar radiation compared to uncovered PV parameters.
modules under the same temperature at STC. As shown

Fig. 2. The I-V, and P-V curves of normal and faulty condition of the PV system at STC.

3
E3S Web of Conferences 107, 02001 (2019) https://doi.org/10.1051/e3sconf/201910702001
ICSREE 2019

In fact, the efficiency of the MPPT device used in the abnormal scenario, they play the main role in the fault
PV system has not the efficiency of 100%, in this paper, detection and diagnosis scenarios.
we assume the MPPT efficiency greater than or equal The output FLC algorithm is based on the
95%). Therefore, these ratios are subjected to high and comparison of the three ratios with the calculated
low limits. threshold high and low limit values, then send an alarm
to the operator for taking correct action as shown in Fig.
4 which illustrates the flowchart of the proposed method.
3.1 High limit OCR VR IR

Where the maximum operating efficiency of the MPPT 3 Crisp inputs

device is applied, therefore, the high limit for OCR, VR Fuzzification

and IR ratios shown in Table 2 is expressed by:

(8) Rules (1-8)
Fuzzy
interface

(9)
Defuzzification

(10) Crisp output

Where , and are the

theoretical parameters of , , and at normal
Fault type (1-8)

conditions respectively at G and T, while

Fig. 3. The Sugeno FLC implementation.
and are the theoretical
data of the same parameters in case of fault conditions at
any G, T respectively. start

Get measured values of Voc,

3.2 Low limit Vm, and Im at solar G, and T

Where the minimum efficiency of the MPPT device Run simulation, and get the simulated values
of Voc, Vm, and Im
is applied. The low limits shown in Table 2 can be
expressed for every case by the following formulas: Calculate the values of OCR, VR,
(11) and IR

(12)
(13) within normal
operating
Yes ) Normal
condition
The difference between this method and the other region

similar fault detection techniques defined in literature is No (Faulty condition)

that this method discusses a number of temporary faults Start FLC algorithm

such as snow cover and bird or tree leaves dropping, also,

it differentiates between the faults with similar features. No ) Other fault
In addition, the proposed fault method can apply for an
within
threshold occurred
unexpected variation of G and T. limits of fault
cases
Yes
Display the detected fault

4 Fuzzy logic control implementation type according to Table 1

) Alarm
FLC is one of the modern artificial techniques used in
fault diagnosis in the PV system. A concise diagram of
Fig. 4. The flowchart of the proposed fault detection method.
the ongoing productions on FL schemes configuration is
displayed in [12]. The FLC system attributes are built Table 2. The overall fuzzy logic system classifier system
based on the condition of if, and then statement between design in normal and faulty conditions
numbers of uncertain input set of data in order to give
decisions called the membership function output. OCR F.
VR limits IR limits
In this paper, Sugeno FL classifier is exhibited and Case PS% Limits Classifier
Min Max Min Max Min Max region
confirmed experimentally for its characteristics which
F0 0 1 1.05 1 1.05 1 1.05 0
have been discussed in [12]. The architecture of the
F1 50 1 1.03 0.97 1.03 1.97 2.11 1
implementation is based on the Max-Min arrangement
F2 50 1.04 1.09 1.03 1.07 2.03 2.14 2
procedure with a centroid type for the defuzzification as F3 0 1.17 1.2 1.16 1.23 1 1.01 3
shown in Fig. 3, moreover, eight FL rules were selected F4 0 1.44 1.48 1.42 1.49 1 1.01 4
and implemented in order to detect accurately the F5 0 1.18 1.25 1.17 1.24 1.01 1.03 5
occurred faults in the PV array as presented in Table 1. F6 0 1.48 1.56 1.49 1.56 1.01 1.04 6
From the previous analysis of the OCR, VR, and IR in F7 0 1 1.15 1.11 1.25 1.08 10 7
the PV system, the fault diagnosis scenarios can be F8 0 1.08 1.14 1.05 1.25 0.97 1.05 8
obtained. Each OCR, VR, and IR region represent a fault
case could happen in the PV system. While , and
have a rapid change during each case of normal and

4
E3S Web of Conferences 107, 02001 (2019) https://doi.org/10.1051/e3sconf/201910702001
ICSREE 2019

5 Results and discussion

5.1 PV system and data collection

The PV array consists of 13 monocrystalline silicon
JKM245 P-60-I PV modules, each module has 60 cells
with 3 bypass diodes (one diode per 20 cells) connected
in parallel with the cells in reverse connection.. Each
module generates 245 Watt output power under STC,
AM1.5 as shown in Table 3. The PV array has 13 Fig. 6. The calculated values of OCR, VR, and IR in different
modules connected in series and one parallel string and fault and normal cases.
installed on the roof of NCEPU, Beijing, China as shown
in Fig. 5 which shows the experimental setup of the PV 5.2 Analysis of the testing FLC results
system. The measurement of power, voltage, and
currents are obtained and collected by internal sensors The membership output of the different fault cases for
every one minute over all day 13th of April 2017. Table 4 Sugeno FL method is shown in Fig. 7. Each examined
shows a full day tested cases at normal radiation and PV condition is categorized in the shown figure. For
temperature which are tested to the PV array. example, case 1 represents F1 fault, for this selected PV
Furthermore, a comparison between the theoretical fault case, the output of the Sugeno fuzzy algorithm is
output vs. the real-time faulted data of PV system during equal to 3. Similarly, case 5 and 6 represents the F7, and
the examined cases of faults are conducted to calculate F8 respectively. In this situation, the output of the fuzzy
the values of OCR, VR, and IR as shown in Fig. 6. algorithm is equal to 7 and 8 respectively.
MPPT efficiency ≥ 95℅ Most of the samples data during the testing time
Indicators and
thresholds period are founded within the low and high threshold
MPPT block in
DC/DC converter
mathematical
calculation
limits. Although, in each case, some samples couldn't be
detected correctly which called out-of-region samples.
Real PV plant
Solar radiation sensor (13 series PV
PV electrical parameters
output
Fault type This state is happening due to the high number of faulty
modules)
Algorithm
regions identified by the fuzzy system and the
MPPT simulation Data analysis and
decision calculation of three ratios may be affected by the sensors
block FLC implementation efficiency. The detection accuracy (DA) for each
Temperature sensor Simulated PV plant examined case can be calculated by:
(14)
Fig. 5. Experimental setup of the fault detection method on the
PV system.

Table 3. Basic parameters for the module at STC condition.

Parameter Variable Value

Maximum power (W) 𝑃𝑃 245
Voltage at max. power (V) 30.1
Current at max. power (A) 8.14
Open circuit voltage (V) 37.5
Fig. 7. The Sugeno fuzzy logic output for the different
Short circuit current (A) 8.76 examined six cases.
Table 4. A full day implemented different cases of normal and Table 5. The output DA without and with the FLC system.
faulty conditions.
Without FLC With FLC
No. of Out of
Start End Condition of PV Case Scenario Out of region DA
case samples DA (%) region
time time system samples (%)
to FLC samples
Two mod. Short 1 F3 8 93.3 1 99.1
1 6:00 7:59 120
circuit F3 2 F6 6 95 2 98.3
Four mod. Open 3 F1 5 95.8 1 99.1
2 8:00 9:59 120
circuit F6 4 F0 4 96.6 0 100
PS w/o Bp. diode 5 F7 5 95.8 1 99.1
3 10:00 11:59 120
failure F1
Normal condition 6 F8 7 94.2 2 98.3
4 12:00 13:59 120
F0 For each case discussed in Table 4, the FL method
Snow falling fault can decrease the out of region samples when compared
5 14:00 15:59 120
F7 with before considering the FLC system as shown in
Bird or tree leaves Table 5. It clearly ensures that most of the samples are
6 16:00 17:59 120
dropping fault F8 classified accurately using the fuzzy classifier. Moreover,
Total=720
the total DA after using FLC system with only out of

5
E3S Web of Conferences 107, 02001 (2019) https://doi.org/10.1051/e3sconf/201910702001
ICSREE 2019

region samples 7 from total samples 720 equals 99% In addition, the FLC is adopted in this study due to
which improves the monitoring of the PV system. As can its straightforwardness and high computational speed,
we see that, the proposed method can precisely and and it can be applicable in case of large scale PV plant
effectively classify, and detect the different faults data. due to low monitoring cost and economic benefits.

5.3 Comparison with other methods References

This step is implemented to check the efficiency of the 1. E. Kaplani, Detection of degradation effects in field-
proposed FLC technique, this evaluation is obtained by aged c-Si solar cells through IR thermography and
comparing with the method in Ref. [11] that ANN to digital image processing, Int. J. of Photovoltaic
identify and detect eight different faults. The comparison Energy, 11, (2012).
between the parameters and the decision output of each
2. A.T. Lahiania, A. B. Abdelghani, I. S. Belkhodja,
method is obtained in Table 6. We can notice that, the
Fault detection and monitoring systems for
ANN model in Ref. [11] can't identify the all presented
photovoltaic installations: A review, Renew. and
fault types (as shown in case 3, Table 6), moreover, it
Sustain. Energy Review, 82, 12 (2018)
has low classification rate which has been implemented
in the confusion matrix the paper compared with the 3. A. Chokor, M. El Asmar, S. V. Lokanath, A Review
proposed method. of PV DC Systems Prognostics and Health Manag.:
The comparison with other method notified that the Challenges and Opportunities, PHM Conf., 6 oct.,
proposed FLC method has a high DA and classification Denver, Colorado, 12(2016).
rate. However, increasing the number of rules using in 4. R. Hariharan, M. Chakkarapani, G. S. Ilango, A
FLC method may lead to more computational time. Also, Method to Detect Photovoltaic Array Faults and PS
the proposed method accuracy may be affected by the in PV Systems, IEEE J. of Photovoltaic, 6(5),
efficiency of the sensors used in measurements. 7(2016).
Table 6. Comparison between the proposed FLC method
5. R. Khenfer, et. al, Faults Detection in a PV
Results and the ANN method in Ref. [11]. Generator by Using Matlab Simulink and the
chipKIT Max32 Board, Int. J. of Photovoltaic
Proposed FLC Energy, 9(2014).
ANN method
Scena method 6. S. Silvestre, et. al., Remote supervision and fault
Case
rio Decisi
Parameters Decision Parameters detection on OPC monitored PV systems, Sol.
on
Algorithm 1:
Energy, 137, 9(2016).
N= 1, C1= 0.01, C2= Group of 7. W. Chine, A. Mellit, A. M. Pavan, S. A. Kalogirou.,
Two 0.01, T1=0.35 A Fault detection method for grid-connected
faults not OCR= 1.18,
1 mod. Algorithm 2:
short
discriminab VR= 1.17, F3 photovoltaic plants". Ren. Energy, 66, 11(2014).
= 0.82, = le IR= 1
circuit 8. R. Platon, J. Martel, N. Woodruff, T. Y. Chau,
0.83
= 0.99
Online fault detection in PV systems". IEEE Trans.
Sustain. Energy, 6(4), 7(2015).
Four Algorithm 1: Open
OCR= 1.48, 9. M. Dhimish, V. Holmes, B. Mehrdadi, M. Dales,
mod. N= 1, C1= 0.01, circuit fault
2 VR= 1.49 , F6
open C2= 0.02, T1=0.35 Ain any cell Fault detection algorithm for GCPV plants, Sol.
IR= 1.01
circuit or (module) Energy., 137, 9(2016).
Algorithm 1: Shadow
OCR= 1.1, 10. M. Dhimish, V. Holmes, B. Mehrdadi, M. Dales,
Snow N= 1, C1= 0.52, with faulty
3
cover C2= 0.5, T1=0.35 A bypass
VR= 1.15, F7 Diagnostic method for photovoltaic systems based
IR= 2 on six layer detection algorithm, Elec. Power Sys.
diode
C1: reduction in the ; C2: reduction or an increase in the ; N: Res. 151, 13(2017).
number of MPPs; T1: Threshold at STC, 11. W. Chine, A. Mellit, V. Lughi , A. Malek, G.
Sulligoi, A. M. Pavan., A novel fault diagnosis
technique for photovoltaic systems based on
6 Conclusion artificial neural networks, Renew. Energy, 90,
In this paper, a new technique for monitoring and 11(2016).
detecting the faults in the PV system is discussed. This 12. M. Dhimish, V. Holmes, B. Mehrdadi, M. Dales,
technique is carried out based on three ratios of the OCR, Comparing Mamdani Sugeno fuzzy logic and RBF
VR, and IR with their thresholds high and low limits. ANN network for PV fault detection, Renew. Energy,
In order to detect failures in the PV system, firstly, 10 (066), 2017.
eight faults were presented in detail to evaluate the 13. S. Das, A. Hazra, M. Basu., Metaheuristic
performance of the algorithm on the PV system. All the optimization based fault diagnosis strategy for solar
experimental faults were accurately identified, classified photovoltaic systems under non-uniform irradiance,
by the proposed method. Moreover, the total DA after Renew. Energy, 118, 15(2017).
using FLC system equals 99% which improves the
monitoring of the PV system.